private void VerifyDocFreq()
{
IndexReader reader = IndexReader.Open(dir);
TermEnum termEnum = null;
// create enumeration of all terms
termEnum = reader.Terms();
// go to the first term (aaa)
termEnum.Next();
// assert that term is 'aaa'
Assert.AreEqual("aaa", termEnum.Term().Text());
Assert.AreEqual(200, termEnum.DocFreq());
// go to the second term (bbb)
termEnum.Next();
// assert that term is 'bbb'
Assert.AreEqual("bbb", termEnum.Term().Text());
Assert.AreEqual(100, termEnum.DocFreq());
termEnum.Close();
// create enumeration of terms after term 'aaa', including 'aaa'
termEnum = reader.Terms(new Term("content", "aaa"));
// assert that term is 'aaa'
Assert.AreEqual("aaa", termEnum.Term().Text());
Assert.AreEqual(200, termEnum.DocFreq());
// go to term 'bbb'
termEnum.Next();
// assert that term is 'bbb'
Assert.AreEqual("bbb", termEnum.Term().Text());
Assert.AreEqual(100, termEnum.DocFreq());
termEnum.Close();
}
public override int DocFreq()
{
if (termEnum == null)
{
return(0);
}
return(termEnum.DocFreq());
}
private void PrintSegment(System.IO.StringWriter out_Renamed, System.String segment)
{
Directory directory = FSDirectory.GetDirectory(indexDir, false);
SegmentReader reader = new SegmentReader(new SegmentInfo(segment, 1, directory));
for (int i = 0; i < reader.NumDocs(); i++)
{
out_Renamed.WriteLine(reader.Document(i));
}
TermEnum tis = reader.Terms();
while (tis.Next())
{
out_Renamed.Write(tis.Term());
out_Renamed.WriteLine(" DF=" + tis.DocFreq());
TermPositions positions = reader.TermPositions(tis.Term());
try
{
while (positions.Next())
{
out_Renamed.Write(" doc=" + positions.Doc());
out_Renamed.Write(" TF=" + positions.Freq());
out_Renamed.Write(" pos=");
out_Renamed.Write(positions.NextPosition());
for (int j = 1; j < positions.Freq(); j++)
{
out_Renamed.Write("," + positions.NextPosition());
}
out_Renamed.WriteLine("");
}
}
finally
{
positions.Close();
}
}
tis.Close();
reader.Close();
directory.Close();
}
internal static void PrintSegment(System.String segment)
{
Directory directory = FSDirectory.GetDirectory("test", false);
SegmentReader reader = new SegmentReader(new SegmentInfo(segment, 1, directory));
for (int i = 0; i < reader.NumDocs(); i++)
{
System.Console.Out.WriteLine(reader.Document(i));
}
TermEnum tis = reader.Terms();
while (tis.Next())
{
System.Console.Out.Write(tis.Term());
System.Console.Out.WriteLine(" DF=" + tis.DocFreq());
TermPositions positions = reader.TermPositions(tis.Term());
try
{
while (positions.Next())
{
System.Console.Out.Write(" doc=" + positions.Doc());
System.Console.Out.Write(" TF=" + positions.Freq());
System.Console.Out.Write(" pos=");
System.Console.Out.Write(positions.NextPosition());
for (int j = 1; j < positions.Freq(); j++)
{
System.Console.Out.Write("," + positions.NextPosition());
}
System.Console.Out.WriteLine("");
}
}
finally
{
positions.Close();
}
}
tis.Close();
reader.Close();
directory.Close();
}
private void PrintSegment(System.IO.StreamWriter out_Renamed, SegmentInfo si)
{
SegmentReader reader = SegmentReader.Get(si);
for (int i = 0; i < reader.NumDocs(); i++)
{
out_Renamed.WriteLine(reader.Document(i));
}
TermEnum tis = reader.Terms();
while (tis.Next())
{
out_Renamed.Write(tis.Term());
out_Renamed.WriteLine(" DF=" + tis.DocFreq());
TermPositions positions = reader.TermPositions(tis.Term());
try
{
while (positions.Next())
{
out_Renamed.Write(" doc=" + positions.Doc());
out_Renamed.Write(" TF=" + positions.Freq());
out_Renamed.Write(" pos=");
out_Renamed.Write(positions.NextPosition());
for (int j = 1; j < positions.Freq(); j++)
{
out_Renamed.Write("," + positions.NextPosition());
}
out_Renamed.WriteLine("");
}
}
finally
{
positions.Close();
}
}
tis.Close();
reader.Close();
}
private void PrintSegment(System.IO.StreamWriter out_Renamed, SegmentInfo si)
{
SegmentReader reader = SegmentReader.Get(true, si, IndexReader.DEFAULT_TERMS_INDEX_DIVISOR, null);
for (int i = 0; i < reader.NumDocs(); i++)
{
out_Renamed.WriteLine(reader.Document(i, null));
}
TermEnum tis = reader.Terms(null);
while (tis.Next(null))
{
out_Renamed.Write(tis.Term);
out_Renamed.WriteLine(" DF=" + tis.DocFreq());
TermPositions positions = reader.TermPositions(tis.Term, null);
try
{
while (positions.Next(null))
{
out_Renamed.Write(" doc=" + positions.Doc);
out_Renamed.Write(" TF=" + positions.Freq);
out_Renamed.Write(" pos=");
out_Renamed.Write(positions.NextPosition(null));
for (int j = 1; j < positions.Freq; j++)
{
out_Renamed.Write("," + positions.NextPosition(null));
}
out_Renamed.WriteLine("");
}
}
finally
{
positions.Close();
}
}
tis.Close();
reader.Close();
}
public override int DocFreq()
{
return(in_Renamed.DocFreq());
}
/// <summary> Test the term index.</summary>
private Status.TermIndexStatus TestTermIndex(SegmentInfo info, SegmentReader reader)
{
Status.TermIndexStatus status = new Status.TermIndexStatus();
try
{
if (infoStream != null)
{
infoStream.Write(" test: terms, freq, prox...");
}
TermEnum termEnum = reader.Terms();
TermPositions termPositions = reader.TermPositions();
// Used only to count up # deleted docs for this term
MySegmentTermDocs myTermDocs = new MySegmentTermDocs(reader);
int maxDoc = reader.MaxDoc();
while (termEnum.Next())
{
status.termCount++;
Term term = termEnum.Term();
int docFreq = termEnum.DocFreq();
termPositions.Seek(term);
int lastDoc = -1;
int freq0 = 0;
status.totFreq += docFreq;
while (termPositions.Next())
{
freq0++;
int doc = termPositions.Doc();
int freq = termPositions.Freq();
if (doc <= lastDoc)
{
throw new System.SystemException("term " + term + ": doc " + doc + " <= lastDoc " + lastDoc);
}
if (doc >= maxDoc)
{
throw new System.SystemException("term " + term + ": doc " + doc + " >= maxDoc " + maxDoc);
}
lastDoc = doc;
if (freq <= 0)
{
throw new System.SystemException("term " + term + ": doc " + doc + ": freq " + freq + " is out of bounds");
}
int lastPos = -1;
status.totPos += freq;
for (int j = 0; j < freq; j++)
{
int pos = termPositions.NextPosition();
if (pos < -1)
{
throw new System.SystemException("term " + term + ": doc " + doc + ": pos " + pos + " is out of bounds");
}
if (pos < lastPos)
{
throw new System.SystemException("term " + term + ": doc " + doc + ": pos " + pos + " < lastPos " + lastPos);
}
}
}
// Now count how many deleted docs occurred in
// this term:
int delCount;
if (reader.HasDeletions())
{
myTermDocs.Seek(term);
while (myTermDocs.Next())
{
}
delCount = myTermDocs.delCount;
}
else
{
delCount = 0;
}
if (freq0 + delCount != docFreq)
{
throw new System.SystemException("term " + term + " docFreq=" + docFreq + " != num docs seen " + freq0 + " + num docs deleted " + delCount);
}
}
Msg("OK [" + status.termCount + " terms; " + status.totFreq + " terms/docs pairs; " + status.totPos + " tokens]");
}
catch (System.Exception e)
{
Msg("ERROR [" + System.Convert.ToString(e.Message) + "]");
status.error = e;
if (infoStream != null)
{
infoStream.WriteLine(e.StackTrace);
}
}
return(status);
}
public static void VerifyEquals(IndexReader r1, IndexReader r2, System.String idField)
{
Assert.AreEqual(r1.NumDocs(), r2.NumDocs());
bool hasDeletes = !(r1.MaxDoc() == r2.MaxDoc() && r1.NumDocs() == r1.MaxDoc());
int[] r2r1 = new int[r2.MaxDoc()]; // r2 id to r1 id mapping
TermDocs termDocs1 = r1.TermDocs();
TermDocs termDocs2 = r2.TermDocs();
// create mapping from id2 space to id2 based on idField
idField = StringHelper.Intern(idField);
TermEnum termEnum = r1.Terms(new Term(idField, ""));
do
{
Term term = termEnum.Term();
if (term == null || (System.Object)term.Field() != (System.Object)idField)
{
break;
}
termDocs1.Seek(termEnum);
if (!termDocs1.Next())
{
// This doc is deleted and wasn't replaced
termDocs2.Seek(termEnum);
Assert.IsFalse(termDocs2.Next());
continue;
}
int id1 = termDocs1.Doc();
Assert.IsFalse(termDocs1.Next());
termDocs2.Seek(termEnum);
Assert.IsTrue(termDocs2.Next());
int id2 = termDocs2.Doc();
Assert.IsFalse(termDocs2.Next());
r2r1[id2] = id1;
// verify stored fields are equivalent
try
{
VerifyEquals(r1.Document(id1), r2.Document(id2));
}
catch (System.Exception t)
{
System.Console.Out.WriteLine("FAILED id=" + term + " id1=" + id1 + " id2=" + id2 + " term=" + term);
System.Console.Out.WriteLine(" d1=" + r1.Document(id1));
System.Console.Out.WriteLine(" d2=" + r2.Document(id2));
throw t;
}
try
{
// verify term vectors are equivalent
VerifyEquals(r1.GetTermFreqVectors(id1), r2.GetTermFreqVectors(id2));
}
catch (System.Exception e)
{
System.Console.Out.WriteLine("FAILED id=" + term + " id1=" + id1 + " id2=" + id2);
TermFreqVector[] tv1 = r1.GetTermFreqVectors(id1);
System.Console.Out.WriteLine(" d1=" + tv1);
if (tv1 != null)
{
for (int i = 0; i < tv1.Length; i++)
{
System.Console.Out.WriteLine(" " + i + ": " + tv1[i]);
}
}
TermFreqVector[] tv2 = r2.GetTermFreqVectors(id2);
System.Console.Out.WriteLine(" d2=" + tv2);
if (tv2 != null)
{
for (int i = 0; i < tv2.Length; i++)
{
System.Console.Out.WriteLine(" " + i + ": " + tv2[i]);
}
}
throw e;
}
}while (termEnum.Next());
termEnum.Close();
// Verify postings
TermEnum termEnum1 = r1.Terms(new Term("", ""));
TermEnum termEnum2 = r2.Terms(new Term("", ""));
// pack both doc and freq into single element for easy sorting
long[] info1 = new long[r1.NumDocs()];
long[] info2 = new long[r2.NumDocs()];
for (; ;)
{
Term term1, term2;
// iterate until we get some docs
int len1;
for (; ;)
{
len1 = 0;
term1 = termEnum1.Term();
if (term1 == null)
{
break;
}
termDocs1.Seek(termEnum1);
while (termDocs1.Next())
{
int d1 = termDocs1.Doc();
int f1 = termDocs1.Freq();
info1[len1] = (((long)d1) << 32) | f1;
len1++;
}
if (len1 > 0)
{
break;
}
if (!termEnum1.Next())
{
break;
}
}
// iterate until we get some docs
int len2;
for (; ;)
{
len2 = 0;
term2 = termEnum2.Term();
if (term2 == null)
{
break;
}
termDocs2.Seek(termEnum2);
while (termDocs2.Next())
{
int d2 = termDocs2.Doc();
int f2 = termDocs2.Freq();
info2[len2] = (((long)r2r1[d2]) << 32) | f2;
len2++;
}
if (len2 > 0)
{
break;
}
if (!termEnum2.Next())
{
break;
}
}
if (!hasDeletes)
{
Assert.AreEqual(termEnum1.DocFreq(), termEnum2.DocFreq());
}
Assert.AreEqual(len1, len2);
if (len1 == 0)
{
break; // no more terms
}
Assert.AreEqual(term1, term2);
// sort info2 to get it into ascending docid
System.Array.Sort(info2, 0, len2 - 0);
// now compare
for (int i = 0; i < len1; i++)
{
Assert.AreEqual(info1[i], info2[i]);
}
termEnum1.Next();
termEnum2.Next();
}
}
/// <summary>Returns true if index is clean, else false.</summary>
public static bool Check(Directory dir, bool doFix)
{
System.Globalization.NumberFormatInfo nf = System.Globalization.CultureInfo.CurrentCulture.NumberFormat;
SegmentInfos sis = new SegmentInfos();
try
{
sis.Read(dir);
}
catch (System.Exception t)
{
out_Renamed.WriteLine("ERROR: could not read any segments file in directory");
out_Renamed.Write(t.StackTrace);
out_Renamed.Flush();
return(false);
}
int numSegments = sis.Count;
System.String segmentsFileName = sis.GetCurrentSegmentFileName();
IndexInput input = null;
try
{
input = dir.OpenInput(segmentsFileName);
}
catch (System.Exception t)
{
out_Renamed.WriteLine("ERROR: could not open segments file in directory");
out_Renamed.Write(t.StackTrace);
out_Renamed.Flush();
return(false);
}
int format = 0;
try
{
format = input.ReadInt();
}
catch (System.Exception t)
{
out_Renamed.WriteLine("ERROR: could not read segment file version in directory");
out_Renamed.Write(t.StackTrace);
out_Renamed.Flush();
return(false);
}
finally
{
if (input != null)
{
input.Close();
}
}
System.String sFormat = "";
bool skip = false;
if (format == SegmentInfos.FORMAT)
{
sFormat = "FORMAT [Lucene Pre-2.1]";
}
if (format == SegmentInfos.FORMAT_LOCKLESS)
{
sFormat = "FORMAT_LOCKLESS [Lucene 2.1]";
}
else if (format == SegmentInfos.FORMAT_SINGLE_NORM_FILE)
{
sFormat = "FORMAT_SINGLE_NORM_FILE [Lucene 2.2]";
}
else if (format == SegmentInfos.FORMAT_SHARED_DOC_STORE)
{
sFormat = "FORMAT_SHARED_DOC_STORE [Lucene 2.3]";
}
else if (format < SegmentInfos.FORMAT_SHARED_DOC_STORE)
{
sFormat = "int=" + format + " [newer version of Lucene than this tool]";
skip = true;
}
else
{
sFormat = format + " [Lucene 1.3 or prior]";
}
out_Renamed.WriteLine("Segments file=" + segmentsFileName + " numSegments=" + numSegments + " version=" + sFormat);
if (skip)
{
out_Renamed.WriteLine("\nERROR: this index appears to be created by a newer version of Lucene than this tool was compiled on; please re-compile this tool on the matching version of Lucene; exiting");
return(false);
}
SegmentInfos newSIS = (SegmentInfos)sis.Clone();
newSIS.Clear();
bool changed = false;
int totLoseDocCount = 0;
int numBadSegments = 0;
for (int i = 0; i < numSegments; i++)
{
SegmentInfo info = sis.Info(i);
out_Renamed.WriteLine(" " + (1 + i) + " of " + numSegments + ": name=" + info.name + " docCount=" + info.docCount);
int toLoseDocCount = info.docCount;
SegmentReader reader = null;
try
{
out_Renamed.WriteLine(" compound=" + info.GetUseCompoundFile());
out_Renamed.WriteLine(" numFiles=" + info.Files().Count);
out_Renamed.WriteLine(String.Format(nf, " size (MB)={0:f}", new Object[] { (info.SizeInBytes() / (1024.0 * 1024.0)) }));
int docStoreOffset = info.GetDocStoreOffset();
if (docStoreOffset != -1)
{
out_Renamed.WriteLine(" docStoreOffset=" + docStoreOffset);
out_Renamed.WriteLine(" docStoreSegment=" + info.GetDocStoreSegment());
out_Renamed.WriteLine(" docStoreIsCompoundFile=" + info.GetDocStoreIsCompoundFile());
}
System.String delFileName = info.GetDelFileName();
if (delFileName == null)
{
out_Renamed.WriteLine(" no deletions");
}
else
{
out_Renamed.WriteLine(" has deletions [delFileName=" + delFileName + "]");
}
out_Renamed.Write(" test: open reader.........");
reader = SegmentReader.Get(info);
int numDocs = reader.NumDocs();
toLoseDocCount = numDocs;
if (reader.HasDeletions())
{
out_Renamed.WriteLine("OK [" + (info.docCount - numDocs) + " deleted docs]");
}
else
{
out_Renamed.WriteLine("OK");
}
out_Renamed.Write(" test: fields, norms.......");
System.Collections.IDictionary fieldNames = (System.Collections.IDictionary)reader.GetFieldNames(IndexReader.FieldOption.ALL);
System.Collections.IEnumerator it = fieldNames.Keys.GetEnumerator();
while (it.MoveNext())
{
System.String fieldName = (System.String)it.Current;
byte[] b = reader.Norms(fieldName);
if (b.Length != info.docCount)
{
throw new System.SystemException("norms for field \"" + fieldName + "\" is length " + b.Length + " != maxDoc " + info.docCount);
}
}
out_Renamed.WriteLine("OK [" + fieldNames.Count + " fields]");
out_Renamed.Write(" test: terms, freq, prox...");
TermEnum termEnum = reader.Terms();
TermPositions termPositions = reader.TermPositions();
// Used only to count up # deleted docs for this
// term
MySegmentTermDocs myTermDocs = new MySegmentTermDocs(reader);
long termCount = 0;
long totFreq = 0;
long totPos = 0;
while (termEnum.Next())
{
termCount++;
Term term = termEnum.Term();
int docFreq = termEnum.DocFreq();
termPositions.Seek(term);
int lastDoc = -1;
int freq0 = 0;
totFreq += docFreq;
while (termPositions.Next())
{
freq0++;
int doc = termPositions.Doc();
int freq = termPositions.Freq();
if (doc <= lastDoc)
{
throw new System.SystemException("term " + term + ": doc " + doc + " < lastDoc " + lastDoc);
}
lastDoc = doc;
if (freq <= 0)
{
throw new System.SystemException("term " + term + ": doc " + doc + ": freq " + freq + " is out of bounds");
}
int lastPos = -1;
totPos += freq;
for (int j = 0; j < freq; j++)
{
int pos = termPositions.NextPosition();
if (pos < 0)
{
throw new System.SystemException("term " + term + ": doc " + doc + ": pos " + pos + " is out of bounds");
}
if (pos <= lastPos)
{
throw new System.SystemException("term " + term + ": doc " + doc + ": pos " + pos + " < lastPos " + lastPos);
}
}
}
// Now count how many deleted docs occurred in
// this term:
int delCount;
if (reader.HasDeletions())
{
myTermDocs.Seek(term);
while (myTermDocs.Next())
{
}
delCount = myTermDocs.delCount;
}
else
{
delCount = 0;
}
if (freq0 + delCount != docFreq)
{
throw new System.SystemException("term " + term + " docFreq=" + docFreq + " != num docs seen " + freq0 + " + num docs deleted " + delCount);
}
}
out_Renamed.WriteLine("OK [" + termCount + " terms; " + totFreq + " terms/docs pairs; " + totPos + " tokens]");
out_Renamed.Write(" test: stored fields.......");
int docCount = 0;
long totFields = 0;
for (int j = 0; j < info.docCount; j++)
{
if (!reader.IsDeleted(j))
{
docCount++;
Document doc = reader.Document(j);
totFields += doc.GetFields().Count;
}
}
if (docCount != reader.NumDocs())
{
throw new System.SystemException("docCount=" + docCount + " but saw " + docCount + " undeleted docs");
}
out_Renamed.WriteLine(String.Format(nf, "OK [{0:d} total field count; avg {1:f} fields per doc]", new Object[] { totFields, (((float)totFields) / docCount) }));
out_Renamed.Write(" test: term vectors........");
int totVectors = 0;
for (int j = 0; j < info.docCount; j++)
{
if (!reader.IsDeleted(j))
{
TermFreqVector[] tfv = reader.GetTermFreqVectors(j);
if (tfv != null)
{
totVectors += tfv.Length;
}
}
}
out_Renamed.WriteLine(String.Format(nf, "OK [{0:d} total vector count; avg {1:f} term/freq vector fields per doc]", new Object[] { totVectors, (((float)totVectors) / docCount) }));
out_Renamed.WriteLine("");
}
catch (System.Exception t)
{
out_Renamed.WriteLine("FAILED");
System.String comment;
if (doFix)
{
comment = "will remove reference to this segment (-fix is specified)";
}
else
{
comment = "would remove reference to this segment (-fix was not specified)";
}
out_Renamed.WriteLine(" WARNING: " + comment + "; full exception:");
out_Renamed.Write(t.StackTrace);
out_Renamed.Flush();
out_Renamed.WriteLine("");
totLoseDocCount += toLoseDocCount;
numBadSegments++;
changed = true;
continue;
}
finally
{
if (reader != null)
{
reader.Close();
}
}
// Keeper
newSIS.Add(info.Clone());
}
if (!changed)
{
out_Renamed.WriteLine("No problems were detected with this index.\n");
return(true);
}
else
{
out_Renamed.WriteLine("WARNING: " + numBadSegments + " broken segments detected");
if (doFix)
{
out_Renamed.WriteLine("WARNING: " + totLoseDocCount + " documents will be lost");
}
else
{
out_Renamed.WriteLine("WARNING: " + totLoseDocCount + " documents would be lost if -fix were specified");
}
out_Renamed.WriteLine();
}
if (doFix)
{
out_Renamed.WriteLine("NOTE: will write new segments file in 5 seconds; this will remove " + totLoseDocCount + " docs from the index. THIS IS YOUR LAST CHANCE TO CTRL+C!");
for (int i = 0; i < 5; i++)
{
try
{
System.Threading.Thread.Sleep(new System.TimeSpan((System.Int64) 10000 * 1000));
}
catch (System.Threading.ThreadInterruptedException)
{
SupportClass.ThreadClass.Current().Interrupt();
i--;
continue;
}
out_Renamed.WriteLine(" " + (5 - i) + "...");
}
out_Renamed.Write("Writing...");
try
{
newSIS.Write(dir);
}
catch (System.Exception t)
{
out_Renamed.WriteLine("FAILED; exiting");
out_Renamed.Write(t.StackTrace);
out_Renamed.Flush();
return(false);
}
out_Renamed.WriteLine("OK");
out_Renamed.WriteLine("Wrote new segments file \"" + newSIS.GetCurrentSegmentFileName() + "\"");
}
else
{
out_Renamed.WriteLine("NOTE: would write new segments file [-fix was not specified]");
}
out_Renamed.WriteLine("");
return(false);
}
public override int DocFreq()
{
return(termEnum.DocFreq());
}